So relays can figure out who the bots are by seeing who doesn't get reply bot spammed?
Only if the relay is getting reply bot spam directly to their relays. Otherwise they could observe false positives.
that seems a good strategy. I'm just using the names and images in the profile, and it's working for now.